As the title indicates, formal logic is used for modeling information retrieval (IR). Readers can expect a literature review (of IR models), supported with graphs, mathematical formulas, and examples that lead to some interesting conclusions.

For those working in the area of IR, it may be a controversial assumption that “logic-based IR models upgrade the IR process from document-query comparison to an inference process,” as the very first IR classical models relied just on logic--for instance, matching query to documents, query formulation, query-document matrix, definition of relevance, and so on. Semantic queries express inferences studied by the IR community without formal logic theory; therefore, its omission is a huge gap of this work. Karen Spärck Jones, a pioneer of term frequency–inverse document frequency (tf-idf), is not even mentioned; Gerard Salton, “the father of IR,” has only one reference.

Another huge gap is the survey structure logic. Instead of developing the concepts of modeling IR, showing the progress of formal logic applications in the area over the years, the article gives mathematical formulas proposed by some researchers and chosen somehow at random. I would appreciate some examples, as well as more information about which models became popular and which went unnoticed.

Readers will first learn some IR terminology in a very generic sense. The potential of this work is presented next, defined as logical models, to help understand relevance, or represent non-logical models or include domain knowledge. Logic-based models are listed with the respective authors: document and query representation, implication-based categorization, uncertainty versus multiple truth values, ranking to uncertainty, and propositional models.

The limited references for the models make it difficult to see which of them have really contributed to IR over the years, and there are no graphs, diagrams, or pseudocode representations. The article may be recommended to mathematicians who like formal logic from a theoretical viewpoint only, without any vision of implementation to IR.